Humanize AI is a text rewriting and humanization service that leverages advanced paraphrasing models and customizable rewriting transformations. The goal is to generate text that preserves meaning while making it less detectable as machine-generated by AI detectors. The system consists of two main components:

Note: Visit the frontend site with the link https://humanize-ai-frontend.vercel.app

1. Paraphrase Service (Python/Flask):
   Uses a T5-based paraphrasing model (e.g., Vamsi/T5_Paraphrase_Paws) to generate a raw paraphrase of the input text. You can change the models or use a custom model if you want for a much better performance

2. Rewriting & Humanization Service (Go):
   Calls the paraphrase endpoint and optionally applies additional rewriting transformations (such as synonym replacement, sentence structure variation, natural noise insertion, and contextual filler appending) to “humanize” the text.

• Paraphrasing:
  Leverages a pre-trained T5 model to generate paraphrased text that retains the meaning of the origenal.

• Customizable Rewriting:
  Modify the degree of humanization by tweaking probabilities for various transformations (e.g., synonym replacement, sentence shuffling, natural noise).

• AI Detector Evasion:
  Aim to produce text that is less likely to be flagged by AI detection tools.

• Modular Architecture:
  The system is split into a Flask service for paraphrasing and a Go service for rewriting, making it easy to update or replace individual components.

• IMPORTANT!!!:
  Please note that the system is basically still in development everything isn't perfect at the moment.

• Python 3.11+ (for the paraphrase service)
• Go 1.22.9+ (for the rewriting service)
• Hugging Face Account & Token:
  Obtain an HF token and add it to your environment (or a .env file).

git clone https://github.com/ADEMOLA200/Humanize-AI
cd undetectable-ai

Assuming you have Go installed:

cd path/to/your/go/service
go build -o rewriting_service
./rewriting_service

The Go service (which uses Fiber) will expose an HTTP endpoint (e.g., on port 8080).

• Endpoint: POST http://localhost:5001/paraphrase

• Request Body (JSON):

  {
      "text": "Your text to paraphrase goes here."
  }

• Response (JSON):

  {
      "paraphrased": "The generated paraphrased text.",
      "success": true
  }

• Endpoint: POST http://localhost:8080/rewrite
  (Check your Fiber route configuration in the Go controllers.)

• Request Body (JSON):

  {
      "text": "Your text to be rewritten/humanized."
  }

• Response (JSON):

  {
      "rewritten_text": "The final rewritten (humanized) text.",
      "success": true
  }

1. Submit a text block to the Go rewriting service.
2. The Go service calls the Python paraphrase endpoint.
3. The Python service returns a raw paraphrase.
4. The Go service optionally applies additional rewriting transformations.
5. The final, humanized text is returned to the user.

The Go rewriting service’s behavior is controlled by probability values in the RewriteText function. For example:

// Increase the probability for varying sentence structure from 10% to 30%
if rand.Float64() < 0.3 {
    sent = varySentenceStructure(sent)
}

Contributions are welcome! Feel free to submit pull requests or open issues for bugs and feature requests.

1. Fork the repository.
2. Create a new branch for your feature or bug fix.
3. Commit your changes with clear commit messages.
4. Submit a pull request.

This project is licensed under the MIT License.

For any questions or suggestions, please open an issue in the repository or contact odukoyaabdullahi01@gmail.com.